|
Load Profiles
Introduction
The electric energy market is an hourly market with
associated bids and hourly spot market prices. Each utility member of
PJM is required to post hourly electric loads for every Load Serving
Entity (LSE) serving
retail load in the utility's zone. To accomplish this, the utility
must estimate hourly loads for customers who do not have hourly
meters.
Most retail customers do
not have meters capable of registering energy usage on an hourly
basis. Load profiling is the process of allocating a customer's
accumulated kWh over a billing cycle to the individual hours in that
cycle. Through load profiling, customers without hourly meters are
able to participate in the electric retail market.
The allocation of total billed energy to specific hours may
be based on historical load usage patterns (static load profiling) or
real-time sample metering (dynamic load profiling). In the initial
estimation for the settlement of energy among LSEs, BGE will use the
static profiling method. In the final estimation and settlement, for
which real-time metering data for the period to be settled have been
collected, BGE will use the dynamic profiling method. Both of these
methods are documented on this web site.
BGE's load profiles are based on average
Historical Hourly Load Data in kWh collected from a statistical sample of the
segment to be profiled. From the sample data an average profile for
each segment is created for each hour in the year. The sample data
used to compute these averages are also utilized to calculate the
hourly weather sensitive load profiles used for the day-after energy
settlement with PJM.
Hourly Weather Sensitive Load Profile Methodology
Annually, a weather-adjusted, average hourly profiled load will be
determined for each profiled segment on a daily basis in accordance
with BGE's load profiling methodology. This methodology is being
implemented in the Load Vision system, which computes profiled loads
using the "Hourly Weather Sensitive" technique. This technique uses a
defined season and day-type structure to run a linear regression of
historical weather data on account load for each account segment. The
profiles created consist of a series of regression equations
expressing the relationship between temperature and load for the
pre-selected season and day-type combinations. The data for these
regressions originate from the 1999 calendar year through the latest
updated calendar year hourly weather and
electric loads from the load research sample for each profiled
segment.
Based on the season/day-type combination selected, Load
Vision generates a weather response function for each hour represented
by the season/day-type combination. The equation relates the
historical loads to values in a temperature range. The linear
relationship is a piece-wise linear regression equation whose
regression parameters are estimated using a search algorithm. The
search algorithm identifies the optimal breakpoints for the regression
lines such that the resulting regression model has the best possible
statistical fit to the historical load data. The algorithm also
ensures that boundary points between adjacent regression line segments
of the weather response function coincide, thereby maintaining a
continuous functional form.
These regression equations are used to compute estimated
load values for the observed temperature points, such that each hour
of the season/day-type combination will have a reported load value
based on that hour's weather. With such a specification, a
representative load for a "typical" electric account within the
profiled segment can be produced. The primary output of this profiling
process is the
load profile:
a set of 24 hourly loads for a day-type and season combination.
BGE has also developed a tool called "BGE Profiler". Users
can input hourly weather data and a date, and BGE Profiler will
calculate static weather sensitive hourly loads for that date. The
tool has been set up in Microsoft ExcelŽ and must be downloaded to
use. It may be found under the
BGE Profiler link on this web
site.
|
